Video To Claude

Name: Video To Claude
Author: AdityaO5

@AdityaO5

关于 Video To Claude

Claude to Dev - easiest way for you showcase close to video issues of your projects, this tool helps you cut out noise and focus only where you need claude to see

基本信息

分类

AI 与智能体

许可证

MIT

运行时

node

传输方式

stdio

发布者

AdityaO5

提交者

Rakesh Aditya

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "video-to-claude": {
      "command": "npx",
      "args": [
        "video-to-claude-mcp"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Video To Claude?

Video To Claude is an MCP server that lets you drop a screen recording, scrub the timeline, draw annotations on frames, and send the annotated WebPs directly into your Claude Code conversation as vision input. It is designed for Claude Code users who need to provide visual context from video.

How to use Video To Claude?

Type /vtc in Claude Code. Option A (recommended) uses the Claude Code plugin: /plugin install github:adityaO5/video-to-claude. Option B installs via npm (claude mcp add video-to-claude --scope user -- npx video-to-claude-mcp). Option C clones and builds from GitHub. After installation, the /vtc command opens a browser; you upload a video, scrub the timeline, annotate frames with red marks, capture frames, and click Send to deliver them to the conversation.

Key features of Video To Claude

Annotate video frames with arrows, boxes, freehand marks, and text labels
Captures annotated frames as WebP images directly into Claude Code
Supports common video formats: .mp4, .mov, .webm, and others
Bundles ffmpeg and sharp – no separate ffmpeg or Python installation needed
Automatic quality retry (80→50) to stay under 2 MB per frame
Slash command /vtc automates capture session and send workflow

Use cases of Video To Claude

Debug UI bugs by circling issues in screen recordings and sending as context
Demonstrate visual changes by annotating before/after frames in a video
Provide visual feedback on design prototypes during code reviews
Send annotated video frames as vision input for Claude to generate or modify code

FAQ from Video To Claude

What are the prerequisites for Video To Claude?

Node.js 20+ (Node 24 recommended), the Claude Code CLI installed, and any of Windows, macOS, or Linux. No Python or separate ffmpeg is needed – both are bundled.

How do I install Video To Claude?

Three options: (A) Claude Code plugin: /plugin install github:adityaO5/video-to-claude; (B) npm: claude mcp add video-to-claude --scope user -- npx video-to-claude-mcp; (C) clone and build from GitHub, then register the MCP server.

What video formats and image constraints apply?

Video: .mp4, .mov, .webm, or any common format. Captured images are WebP, target ≤2 MB per frame (hard cap 5 MB), max width 960 px. Practical limit ~20 frames before auto-downscale; soft cap ~100 frames.

How does the MCP server work with Claude Code?

It exposes two tools: start_capture_session (starts dev server, creates session, opens browser) and await_capture (polls until user clicks Send, returns annotated WebPs). The /vtc slash command calls both in sequence automatically.